True Partner Mike Montano built a Mac app on a 45-minute flight home. Custom software isn't just for big teams anymore because the math on build vs. buy is shifting. 

I've noticed something remarkable lately. I keep building software that wouldn't have made any sense to build a few years ago. Not side projects. Stuff I use every day.

A few weeks ago, on a 45-minute flight home, I had Claude build a Mac app for the software we built to replace the tool we’d used for a decade. By the time I landed, the app existed. More and more, I can build about as fast as I can think.

Custom software used to be expensive to build, host, secure, and keep running. So you bought the generic product and bent your work around it. I think that math is breaking. More and more it's just easier to build the exact thing I need than to jam a square peg into a round hole.

What we've built

We used project management software to coordinate our work  at True for about a decade. It's good software, but we have weird workflows. So we rebuilt it. Keerthi Nalabotu did the first pass in a few days, and now we have Truebase - software built exactly to our weird workflows - and we love it. The flow is all about markets, deal flow, portfolio, and our network, and it sits on top of our data lake. So when a new company comes in, there's an agent right there that can pull the entire thread: the 20 similar companies we've seen over the last two decades, why we passed, what we wanted to see and never did. And our team isn't just people anymore. Agents are part of the work now, and the tool we were using wasn't built for that. Our old software was never going to do that for us.

Then we decided to build our own messaging app, which sounds a little crazy. Messaging is about as solved as software gets. But when a few of us swarm on a deal, we want that conversation tied to the company, visible to the right people, and sitting next to everything else we know. So we built native iPhone and Mac apps on the same backend as Truebase.

We’re not building for the sake of it. Our job is meeting founders and picking great deals. Custom software is increasingly how we free up time to spend more time with extraordinary builders.

Same thing at home. Family life has plenty of its own moving parts, so I built my own version of Instinct. Mine is more useful to me because it can see everything: my email, messages, finances, and the agents and integrations I'd already built. Several things I'd never give a generic app access to. My own software can have it, and the raw data stays on my machine.

Plus smaller stuff: a workout player that syncs my intervals, Spotify, and coaching cues. I've also built apps to help friends run their small businesses.

I don't just want chat

I don’t think the endgame is to just ask the AI for everything. I want intelligence and UX in the right combination at the right moment, depending on what I'm doing, where I am, and what I need.

A lot of the time I want to see something, not read about it. If I ask what my workouts look like for the week, I want to see it laid out. If I ask what needs attention at home, I want a list I can act on. Built right when I need it, for exactly what I asked. I think that's going to show up inside the AI tools we already use. We have an early version of this: a deal cockpit that runs inside Claude Code.

It'll be hard to get right. Nobody can learn a new UX every 30 minutes, so it has to feel familiar. And some things need to be deterministic. When I approve something, I want it to work exactly the same way every time, not however the model feels like doing it today. Intelligence plus predictable software.

The way I build has changed too. Almost everything now has two front doors from day one: an app for people, and an MCP server or API for agents. Truebase has both. My home infrastructure has both. And picking between web, iPhone, and Mac used to be a real tradeoff. Now I just build all of them.

Build it, use it, throw it away

Some software won't need to last. For our Connected Stack conference, we built an app just for the people attending: the schedule, who to meet, and a way to message each other. It only had to work for one event.

And some software will keep changing after it ships. Something like a software factory, where the app keeps building itself and personalizing for each user. We’re starting to do this with Truebase, so it fits how each person on our team actually works. What ships on day one is just the starting point.

More software, not less

I'm not saying build everything. There are plenty of hard problems I'm happy to pay someone else to solve. But the line between build and buy is moving fast.

That Mac app from the flight will only ever have a handful of users. A few years ago that would have made it a waste of time. Now it's just how I work.

Some people will hear "a lot more software" and assume a lot more slop. Maybe some of it will be. But I think most of it will be genuinely useful software for specific jobs and small groups of people. Sometimes as small as one person, for one moment in time.